About us

An AI research lab, built where the problems are.

We started with a simple observation: the places with the most to gain from AI are the places it reaches last. Not because the technology can't help, but because it wasn't designed to. So we're designing it differently: private, close, open, and within reach.

Why we exist

About the lab.

Oragonlabs started with engineers building AI for institutions that needed it, and watching the work fall over for reasons that had nothing to do with the models. The models were fine. Everything around them took things for granted: a data centre nearby, a network that stayed up, and an organisation happy to send its records abroad just to get an answer back.

Two things got in the way, and neither of them was the technology. One was what it cost to even start. Using AI meant rebuilding: new infrastructure, new specialists, money signed off long before anyone had seen it work. The other was trust. No hospital is going to hand over patient records, and no ministry is going to hand over citizen data, to a system it can't inspect, running somewhere it can't visit, under terms it didn't write. Most of them looked at that and said no.

We're a research lab before we're a product company. Our work starts with a question: can this model run on hardware this institution already owns, in the language its users actually speak, without its data ever leaving the building? It ends with a system somebody depends on.

Research first

We publish what we learn. Field notes, benchmarks and failure reports, not just launch announcements.

Applied, not abstract

Every research direction is chosen because a real institution has the problem today, not because it is fashionable.

Local by construction

We build with teams who live in the markets we serve, in the languages people actually speak.

The founding team

The people building it.

Proximity is a pillar, so who we are and where we sit is part of the argument, not a footnote.

David Azemoh

Founder · Chief Engineer & AI Researcher

9+ years

Nine years architecting distributed systems serving more than 5 million monthly active users, across national infrastructure, regulated finance and public-sector platforms.
Led engineering at Treepz and Thrive Agric, on products driving more than $200M in value.
Two-time founder with an exit and an NVIDIA Inception founder, who has helped startups raise $56M. Certified ethical hacker.
Researches how models actually get deployed: inference that holds under real load, systems that keep working offline, and what it takes to run them on hardware an institution already owns.
Builds and leads the engineering teams that keep those systems running.

AI infrastructureEdge deployment

Daniel Adama

Founder · Chief AI Researcher

8+ years

Spearheaded George, an AI auditor built to ISA, IAASB and GAAS.
Built a multi-tenant AI reconciliation system for banks and fintechs that matches 220,000+ transactions against the general ledger in under 12 minutes, on statement data that arrives messy and inconsistent.
Fine-tuned YOLOv8 across 142,000+ images and 14 classes for real-time detection: cameras placed, data annotated, variants benchmarked, model shipped.
Fine-tuned and deployed an open-source LLM end to end in early 2023, gathering and processing the training data himself.
Builds for what compute costs: serverless GPUs for the heavy vision jobs, a cheap model for bulk matching with heavier processing held back for ambiguous cases, event-driven workers that scale to zero between runs.

Applied researchProduction AI at scale

Between them, more than 15 years building full-stack, reliable systems that have served over 200 million users and generated value worth more than $1B.

Hiring: we're looking for research and deployment engineers. Introduce yourself.

Come build with us.

Partners, researchers and engineers: if this is the lab you've been waiting for, say hello.

Get in touch: hello@oragonlabs.com